Is Artificial Intelligence Privatizing Knowledge?

National Library of Latvia. Photo Courtesy of Shutterstock.

Like money, real estate, and fine art, knowledge can derive value from scarcity.

Despite globalization’s myth of an “Open Society”– the term coined by philosopher Karl Popper which George Soros’ foundation derives its name from— knowledge is not inherently democratic in form, and certain types of knowledge exist only in non-public spaces. The Vatican’s library contains unpublished manuscripts and artifacts which likely contain valuable insight into history, and are only accessible to the elite of this one religious order. Data brokers sell information which has value to the organizations buying it, and NDAs are put in place to protect trade secrets.

As AI continues its deployment to as many users as possible in what has been described as an “AI arms race,” knowledge finds itself increasingly privatized, with many different types of knowledge generated through data collection, and expanded upon via AI algorithms to create new sets. AI systems can analyze vast amounts of data and generate new insights, which is kept private to protect the competitive advantage of those who possess it. AI models, for instance, that analyze financial data create new trading strategies that are highly valuable to financial firms.

Social media companies in the United States “democratized” access to certain types of information, while privatizing other types (the data generated by its users), and created a marketplace for this information; a marketplace restricted from the broader public which prevents knowledge from being used for the public good. Search engines like Google appear to democratize knowledge by making publicly available historical accounts (books, news articles, videos), but function as information exchanges wherein new understandings of history are being developed in real-time only for those with access to it.

AI is now deployed in these private spaces to gain knowledge from these trends, increasingly driving knowledge into the hands of a concentrated few. 

“With the print fixation of the traditional archival terminology, we run the risk of overlooking the fact that a different kind of archive is being built in nonpublic, proprietary ways by entrepreneurs like Bill Gates with Corbis image bank, which holds the digital copyright on a lot of European historical imagery,” writes the media theorist Wolfgang Ernst in Digital Memory and the Archive. “Data trash is, positively, the future ground for media-archeological excavations.

The cliché goes that “knowledge is power,” a quote going back as far as Francis Bacon in 1597. Accordingly, power is found through competitive advantages over other human beings, in which knowledge is subverted into a mechanism for this end, and AI plays the part of (willing) accomplice.